Vol. 175, No. 14

Resolution of Holliday Intermediates in Recombination andDNA Repair: Indirect Suppression of ruvA, ruvB, and

ruvC MutationsTIKSHNA N. MANDAL, AKEEL A. MAHDI, GARY J. SHARPLES, AND ROBERT G. LLOYD*

Department of Genetics, University ofNottingham, Queens Medical Centre,Nottingham NG7 2UH, United KingdomReceived 8 March 1993/Accepted 13 May 1993

The ruvA, ruvB, and ruvC genes of Escherichia cofi provide activities that catalyze branch migration andresolution of Holliday junction intermediates in recombination. Mutation of any one of these genes interfereswith recombination and reduces the ability of the cell to repair damage to DNA. A suppressor ofruv mutationswas identified on the basis of its ability to restore resistance to mitomycin and UV light and to allow normallevels of recombination in a recBC sbcBC strain carrying a TnlO insertion in ruvA. The mutation responsiblewas located at 12.5 min on the genetic map and defines a new locus which has been designated rus. The russuppressor works just as well in recBC sbcA and recm sbc' backgrounds and is not allele specific. Mutationsin ruvB and ruvC are suppressed to an intermediate level, except when ruvA is also inactive, in which casesuppression is complete. In all cases, suppression depends on RecG protein, a DNA-dependent ATPase thatcatalyzes branch migration of Hollidayjunctions. The rus mutation activates an additional factor that probablyworks with RecG to process Holliday junction intermediates independently of the RuvAB and RuvC proteins.The possibility that this additional factor is a junction-specific resolvase is discussed.

Damage to DNA is unavoidable, and all organisms haveevolved enzymatic systems to promote repair and limitmutation. Repair of UV-damaged DNA has received partic-ular attention, and the mechanisms involved have beenstudied in some detail in Escherchia coli, where a largenumber of genes have been identified and their productshave been purified and characterized in vitro. Most UV-

induced lesions are removed by nucleotide excision repairworking in conjunction with enzymes that catalyze homolo-gous recombination. Excision repair removes a broad spec-trum of lesions in a multistep reaction catalyzed by the Uvrproteins, DNA polymerase I, and DNA ligase (46). Thecombined action of the UvrA, UvrB, and UvrC proteinsleads to incision of the damaged strand on either side of thelesion. The 12- to 13-nucleotide fragment containing thelesion is then released by the helicase activity of UvrD, thegap formed is filled by DNA polymerase I (Pol I), and thenew patch is sealed by ligase.The enzymes of recombination come into play when the

replisome encounters a lesion in the template strand andDNA synthesis comes to an abrupt halt. RecA protein allowsthe cell to recover from this situation by helping the repli-some to resume DNA synthesis (7). Part of this process ofrecovery involves recombinational exchanges with the un-damaged sister duplex (9, 32, 33). One possible mechanism isbased on the ideas proposed originally by Howard-Flandersand colleagues and modified subsequently to take into ac-count the ability of RecA protein to catalyze pairing andstrand exchange reactions between homologous DNA mol-ecules (10, 48, 49). According to this model (Fig. 1), DNAsynthesis resumes downstream of the lesion via a mecha-nism that remains uncertain but which leaves a gap oppositethe lesion in the template strand. RecA polymerizes at thegap to form a nucleoprotein filament that promotes homolo-

* Corresponding author.

gous pairing and strand exchange with the undamaged sistermolecule. Strand transfer past the lesion closes the gap andallows the excision enzymes a further opportunity to repairthe DNA. Further extension into duplex regions may resultin reciprocal exchange and the formation of a classicalHolliday junction (8). Endonuclease cleavage across thepoint of strand exchange resolves the connection betweenthe sister molecules, leaving DNA polymerase to fill in thegap formed in the parental strand. Otherwise, the moleculescould be separated by reversing the direction of branchmigration. An alternative mechanism for bypassing the le-sion is for the replisome to switch to copying the undamageddaughter strand and to switch back to the parental strandafter clearing the lesion. In this model, the recombinationcomplex formed by RecA simply provides the means toswitch strands (7).Although the mechanism(s) of recombinational bypass

remains to be established, recent studies have identifiedactivities encoded by the ruv and recG genes that couldprocess the intermediates generated by RecA into viableproducts. As with recA, mutations in these genes confersensitivity to UV light, particularly in an excision-deficientbackground where lesions persist and have to be bypassed atevery round of replication (9, 18, 20, 29). Three genes havebeen identified at the ruv locus (36). The ruvA and ruvBgenes encode proteins of 24 and 37 kDa, respectively, thatact together to catalyze branch migration of Holliday junc-tions. RuvA targets RuvB to the junction where the ATPaseactivity of the latter provides the motor to drive branchmigration (11, 12, 31, 38, 44, 45). A third ruv gene, ruvC,encodes a 19-kDa nuclease that resolves junctions by sym-metrical strand cleavage across the point of strand exchange(5, 6, 13).The sequential action of RecA, RuvAB, and RuvC pro-

vides what appears to be a simple enzymatic pathway for therepair of strand gaps. However, the situation is complicatedby a functional overlap between ruv and recG. The recG

4325

JOURNAL OF BACTERIOLOGY, July 1993, p. 4325-43340021-9193/93/144325-10$02.00/0

on Septem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

4326 MANDAL ET AL.

(i)

Homologous pairino and Branch miaration Excision and ranair_,-r A--. -.--vna%"g LAdIFW 1 421 GusI tral * wlut" UY CaudyaquReplication strand exchange synthesis

(ii)A

(iii) A (v) UvrABC (V) (vi), RecA -------- RuvAB -R UvrD -# RuvC ..

3-00 P. ..& -PoI X _ XRecG ~~t Io

Holliday junction RecG RuvAB(vii)(vii) - Resolution by

reverse branch- migration

FIG. 1. A model for postreplication repair of UV-damaged DNA. A DNA replication fork encounters a UV-induced pyrimidine dimer(closed triangle) in the template strand (solid line) (i) and resumes DNA synthesis downstream, leaving a gap in the daughter strand (shadedline) (ii). RecA protein polymerizes at the gap and initiates strand exchange with the sister duplex (iii). Branch migration by RuvAB or RecGextends the heteroduplex joint beyond the lesion and leads to a Holliday junction (iv). Excision of the dimer (v) followed by further processingof the junction by RuvC cleavage (vi) or by reverse branch migration with RecG or RuvAB (vii) separates the sister duplexes and allows repairto be completed by filling in the remaining gaps.

locus encodes a 76-kDa protein which has ATPase andDNA-binding activities. It interacts specifically with syn-.thetic Holliday junctions and dissociates these to duplexproducts indistinguishable from those produced by RuvAB(25). It also catalyzes branch migration of Holliday interme-diates made by RecA in vitro (51).The overlap between RuvAB and RecG is consistent with

the fact that both ruv and recG single mutants produce some30 to 50% of the normal yield of recombinants in geneticcrosses whereas ruv recG double mutants produce 0.5% orless (17). Presumably, these proteins provide equally effi-cient alternative activities for producing recombinants.However, the same cannot be said for repair of damagedDNA since ruv mutants are far more sensitive to UV lightthan recG strains. The reason for this difference betweenrecombination and DNA repair is not clear. The ruvA andruvB genes form an operon regulated by LexA protein andare induced in response to DNA damage (4, 39, 40). recG isa component of the spoT operon and is not SOS inducible(14, 24). Perhaps UV-irradiated ruvAB mutants have insuf-ficient RecG to cope with all of the lesions generated. Asecond problem arises from the fact that ruvC recG doublemutants have the same extreme sensitivity to UV light andsevere deficiency in recombination as ruvA recG or ruvBrecG strains (17). There is no evidence that RecG has anuclease activity that could cleave junctions in the absenceof RuvC (25).

In this article, we describe a suppressor of ruv mutationsthat allows both recombination and DNA repair to proceedindependently not only of RuvAB but also of RuvC. Theproperties of the suppressor raise the possibility that ruvA,ruvB, and ruvC mutants are defective for resolution ofHolliday junctions and that branch migration proceedslargely unhindered in these strains via RecG. The datapresented support models for recombinational bypass ofUVlesions that require resolution of Holliday junctions.

MATERIALS AND METHODS

Strains. E. coli strains are listed in Table 1. The AruvA63allele in N2096 has the same polar effect on ruvB as theoriginal ruvA60::TnlO insertion (36). XNK55 and X1105 are

transposon hop vectors for generating mini-kan insertions(41, 47).

Plasmids. pBL125 and pBL135 are recG' derivatives ofpBR322 and pTZ18R, respectively (24). pGS751 is a ruvC'derivative of pGEM-7Zf(+) (37). pNK2859 was used toprovide the 1.7-kb BamHI probe for the kan insertiongenerated with X1105 (15). pRS415 is a pBR322-based vectorfor inserting promoters in front of the lacZYA genes (43).pTM101 and pTM102 were made by cloning the 4.1-kbEcoRI-PvuII and 4.5-kb EcoRI-StuI DNA fragments, re-spectively, from the recG promoter region in pBL125 intopRS415 cut with EcoRI and SmaI.DNAs. Synthetic X-junction DNA was made by annealing

the oligonucleotides (49- to 51-mer) numbered 1, 2, 3, and 4described previously (25, 30). Oligonucleotide 1 was 3pend-labelled at the 5' end before being annealed. DNAconcentrations are in nucleotide equivalents. Junction DNAwas measured with DNA DipSticks (Invitrogen, San Diego,Calif.), and the measurement is approximate because of thelow concentration.

Proteins. E. coli RuvA, RuvB, and RecG proteins werepurified as described previously (25, 44).Media and general methods. LB broth and 56/2 salts media

have been described (22). The LB medium contained 0.5 g ofNaCl per liter except for matings, for which the salt concen-tration was increased to 10 g/liter. Broth and agar mediawere supplemented with 20 ,ug of tetracycline per ml, 40 pugof kanamycin per ml, or 50 pg of ampicillin per ml, asrequired for strains carrying antibiotic-resistant plasmids ortransposons. Plasmid transformations and methods for mea-suring sensitivity to UV light have been described before(19, 23, 36). UV irradiation was delivered at a dose rate of 1J/m2/s. Plate tests for measuring sensitivity to mitomycinused LB agar containing mitomycin at concentrations ofeither 0.2 or 0.5 jxg/ml. Media and methods for propagatingphage X and for selecting tetracycline-sensitive isolates ofstrains carrying TnlO insertions followed the recipes andprotocols of Silhavy et al. (42). Assays for P-galactosidasewere as described by Miller (27). Methods for analysis ofplasmid and chromosomal DNA followed recipes and proto-cols described by Sambrook et al. (34).

J. BACTERIOL.

Rasqnlijtinn hv t-lanunnos

on Septem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

INDIRECT SUPPRESSION OF ruv MUTATIONS 4327

TABLE 1. Escherichia coli K-12 strains

Strain Relevant genotypea Other Source or reference

IN(rrnD-rrnE)lrec + ruv+ s+recB21 recC22 sbcBIS sbcC201AspoVl::kan (= ArecG263)pyrE60adk-2ruvC53 eda-Sl::TniOAruvC64::kanA(lac-pro)XIIIruvA60::TniOAruvA63ruvA60 recBC sbcBCrecG258::kanrecGI62purE85::TniOIN(rmD-rmE)Ieda-SI IN(rmD-rmE)IpurE85zbc-2252::kan IN(rrnD-rmE)Izbc-2253::kan IN(rrnD-rrnE)Irus-I AruvA63 purE85 zbc-2252rus-i zbc-2252 IN(rrnD-rmE)IruvA60 AruvC64rus-i ruvA60 recBC sbcBCA(ruvA-ruvC)65ruvA60 recGi62AmuvAC65 eda-SIrus-I sbcC rpoB his+ Tcsrus-I purE85 sbcCrus-i purE85 AruvA63recBC sbcBCpurE85rus-irus-i recBC sbcBCrus-i ruvA60rus-i ruvC53 eda-SJrus-i recBC sbcBC ruvC53 eda-SJrecBC sbcBC ruvC53 eda-SJrus-i ruvA60 recBC sbcBCrus-i recG258ruvA60 recG258rus-i ruvA60 recG258adk-2purE85ArecG263rus-i purE85 A(lac-pro)xj11purE85 A(lac-pro)xIIjmus-i ruvA60 AruvC64mvA60 AmvC64 recBC sbcBCrus-i AruvC64pyrE60 ms-i zbc-2252mus-i ruvA60 AmvC64 recBC sbcBCms-i AmvAC65 eda-SIpyrE60 ms-i zbc-2252 AruvAC65 eda-SIrecG263 ms-i AruvAC65 eda-5IAruvAC65 recG263 eda-SIrecG162 ms-i zbc-2252 AmvAC65 eda-SIF' (F128) lacI3 lacZi18 proAB+Hfr (Hayes, P01) A(proB-lac)xII1Hfr (Cavalli, PO2A) reLI tonA22Hfr (PO3 of P4X) reU4I tonA22Hfr (H, P01) (Aind)+ thi-I relAI

AA

ABAAC

AAAAA

A

A

AAAAAAAAAAAAAAAAAAABAC

C

AAAAAAAAAAD

14CGSC 4518cJ. Cronan (CGSC 4682)41This workThis work41Tcs selection on N2057PLN2057 x JC7623 to Tcr2020XNK55 x W3110 to Kmr41PLN3005 x AB1157 to TCrX1105 x W3110 to KmrX1105 x W3110 to KmrPLN3603 x TNM716 to KMrPLN3608 x W3110 to KmrPLGS1481 x N2057 to Kmr (Tcr)This workN3684 to Tcs (and KmS)P1.N2057 x N2973 to TCrP1.N3041 x AM547 to TCrd

PLN3105 x TNM482 to TCrP1.TNM668 x N2096 to TCrP1.N3105 x JC7623 to TCrP1.TNM482 x N3105 to Pur+P1.TNM482 x TNM718 to Pur+PLN2057 x TNM733 to TCrP1.CS85 x TNM733 to TCrP1.CS85 x TNM734 to TCrP1.CS85 x JC7623 to TCrP1.N2057 x TNM734 to TCrP1.N2731 x TNM733 to KmrPLN2731 x N2057 to KmrPLN2731 x TNM759 to KmrPLN3105 x CV2 to TCrPLCF3324 x AB1157 to KmrP1.TNM716 x N1585 to TcTP1.TNM716 x N1585 to TCrPLN3684 x TNM733 to TCrPLN3684 x JC7623 to TCrP1.GS1481 x TNM733 to KmrP1.N3617 x AT2538 to KmrP1.N3684 x TNM734 to TCrP1.AM561 x TNM733 to TCrP1.AM561 x TNM1194 to TCrP1.CF3324 x TNM1208 to KmrP1.CF3324 x AM561 to KmrP1.N2973 x TNM1209 to Pyr+K. B. LowCGSC 5263CK. B. LowK. B. Low (CGSC 4515C)R. Devoret

a After the first full listing, recB21 recC22 sbcBl5 sbcC201 is abbreviated to recBC sbcBC, A(ruvA-ruvC)65 is abbreviated to AruvAC65, and insertions are

abbreviated to the gene symbol plus allele number.b A, F- thi-1 his-4 A(gpt-proA)62 argE3 thr-l leuB6 kdgK51 rfbDI(?)ara-14 lacYI galK2 xyl-S mtl-I tsx-33 supE44 rpsL31; B, Hfr (PO2A) tonA22 AphoA8

ompF627fadL701 relAI glpR2glpD3 pit-10 spoTI; C, same as that for A except thr+ ara+ leu'A(lac-pro)xjj1 (N1585 was made by mating AB1157 topro+ lac+with Hfr KL226 and then to thr+ ara+ leu+ A(pro-lac)x111 with Hfr 30OOxOII); D, A(pro-Iac)x1jj recAl rpsE xyl mtl.

I Strain supplied from the E. coli Genetic Stock Center (CGSC) by B. J. Bachmann.d -, Made by transducing a thyA derivative of RG108 to his' sbcB+ and then to thyA+ recBC+ with P1.W3110 before selecting for resistance to rifampin (100

pg/ml) and sensitivity to tetracycline.

W3110AB1157JC7623CF3324AT2538CV2CS85GS1481N1585N2057N2096N2238N2731N2973N3005N3041N3105N3603N3605N3608N3617N3684RG108AM547AM549AM561TNM482TNM668TNM716TNM718TNM733TNM734TNM759TNM777TNM838TNM839TNM844TNM883TNM896TNM897TNM1031TNM1072TNM1106TNM1107TNM1115TNM1120TNM1124TNM1194TNM1206TNM1208TNM1209TNM1215TNM1219TNM1228KL5483000xL26KL226KL227GY2200

VOL. 175, 1993

on Septem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

4328 MANDAL ET AL.

Genetic crosses and measures of recombination. F' and Hfrdonors were mated with F- recipients in high-salt LB brothat 370C by using procedures described in detail elsewhere(21, 23). Measures of cell viability relate to the number ofCFU in the recipient cultures at an A650 of 0.4. Transcon-jugants were selected on 56/2 or LB agar, as appropriate,supplemented with 100 pg of streptomycin per ml to coun-terselect donor cells. Transductions with phage Plvir fol-lowed the recipes and protocols described by Miller (27).

Isolation of strain RG108. The ruvA60 recBC sbcBC strain,N2238, was grown overnight in LB broth, and 0.05-mlsamples were spread on LB agar containing tetracycline tomaintain selection for TnlO and mitomycin (0.2 pg/ml) toselect for revertants. The plates were then irradiated at 30J/m2 before being incubated overnight at 37°C. Colonies ofmitomycin-resistant survivors were purified and subjected tofurther tests. RG108 was identified by its resistance to bothUV light and mitomycin.

Insertion of mini-kan near purE. Random insertions ofmini-kan were generated in strain W3110 by using X1105 andthe media and protocols of Kleckner et al. (15). To identifyinsertions near purE, P1 phage grown on the pooled Kmrcolonies was then used to transduce strain N3005 to Pur' onglucose minimal agar supplemented with kanamycin.

Deletion of the chromosomal ruvC gene and construction ofa A(ruvA-ruvC) strain. pGS751 was digested with NdeI andStuI, and after the NdeI ends were filled, the larger fragmentwas ligated to a 1.2-kb SmaI kan DNA fragment frompUC4-KIXX (Pharmacia) before being transformed intoJC7623. Kmr transformants were selected in the absence ofampicillin to allow segregation of plasmid-free cells as de-scribed by Oden et al. (28). Recombinants carrying theAruvC::kan allele (AruvC64) inserted in the chromosomewere identified by their sensitivity to mitomycin and UVlight. The mutation was then transduced to AB1157 to givestrain GS1481. The AruvC64::kan genotype was confirmedby Southern analysis (data not shown). AruvC64 was trans-duced into strain N2057 to give the ruvA60::TnlO ArvC64strain N3684, which was identified as a rare transductantresistant to both kanamycin and tetracycline. Deletion of theruvA-ruvC region was achieved by selecting tetracycline-sensitive isolates of N3684 and screening for those clonesthat had lost resistance to kanamycin. One such isolate(AM547) was shown by Southern analysis (data not shown)to be deleted for most, but not all, of the kan sequences. Weassume that this deletion, designated AruvAC65, has re-moved the ruvA sequences upstream of the original TnlOinsertion. Complementation studies with cloned ruv genesconfirmed the ruvAC genotype of AM547. They also re-vealed some expression of ruvB but no more than that seenpreviously in a ruvA60::TnlO strain (36).

Gel retardation assay. Reaction mixtures (20 pl) contained32P-labelled synthetic Holliday junction or linear duplexDNA (-0.15 pM) in binding buffer (50 mM Tris-HCl [pH8.0], 5 mM EDTA, 1 mM dithiothreitol, 100 ,ug of bovineserum albumin per ml) and various amounts of protein. Theproteins were mixed into the reaction mixtures before junc-tion DNA was added. After 15 min on ice, 5 pl of loadingbuffer (50 mM Tris-HCl [pH 7.5], 4 mM EDTA, 25%glycerol, 400 jig of bovine serum albumin per ml) was added,and the samples were loaded immediately onto 4% poly-acrylamide gels in low-ionic-strength buffer (6.7 mM Tris-HCl [pH 8.0], 3.3 mM sodium acetate, 2 mM EDTA).Electrophoresis was carried out at room temperature for1.75 h at 200 V with continuous circulation of buffer. Gelswere dried on Whatman 3MM paper and autoradiographed.

Dissociation of a synthetic Holliday junction. Reactionmixtures (20 A.l) contained synthetic Holliday junction (-0.4p.M) in reaction buffer (50 mM Tris-HCl [pH 8.0], 1 mMdithiothreitol, 5 mM MgCl2, 100 pg of bovine serum albuminper ml, 5 mM ATP) and various amounts of protein. Reac-tions were initiated by the addition of RecG. Reactionmixtures were incubated at 370C for 30 min before theaddition of 5 Atl of stop buffer (2.5% [wt/vol] sodium dodecylsulfate, 200 mM EDTA, 10 mg of proteinase K per ml) andthen incubated for a further' 10 min at 370C. The DNAproducts were then electrophoresed at room temperaturethrough a 10% native polyacrylamide gel by using a Tris-borate buffer system (30) and subsequently autoradio-graphed.

RESULTS

Indirect suppression ofa ruvA::TnlO allele. Strains carryingthe ruvA60::TniO allele give rise to the occasional revertantwith increased resistance to UV light and mitomycin. Theseare particularly noticeable in a recBC sbcBC backgroundwhere ruv mutations confer a more extreme phenotype (18).Strain RG108 is typical of the revertants encountered. It is asresistant to UV light as its ruv+ ancestor, JC7623, andproduces about 50% as many Pro' recombinants in crosseswith Hfr KL227. Its immediate ruvA60 parent strain, N2238,is by comparison very sensitive to UV light and producesrecombinants at a 500-fold-lower frequency than JC7623(Fig. 2a and data not shown). Backcrosses to strains AB1157and JC7623 confirmed that RG108 retains the TniO insertionin ruvA in an unaltered form (data not shown). We concludethat its resistance to DNA-damaging agents is due to anadditional mutation which we designate mus-i for ruv sup-pression. The suppression is specific to ruv mutations. Nosuppression was detected with mutations in any of the othergenes known to affect recombination (data not shown).

Location of the suppressor. Genetic crosses (data notshown) linked rus- to purE, which is at 12.2 min on thegenetic map (Fig. 3a). rus- single mutants segregating inthese crosses were found to be slightly sensitive to mitomy-cin at 0.5 pug/ml (data not shown). However, they areresistant to UV light (Fig. 2c). To locate rus more precisely,we isolated mini-kan insertions linked to purE and posi-tioned these relative to the Kohara et al. physical map (16)by Southern analysis of chromosomal DNA digests. Theprobe used was the 1.7-kb BamHI DNA fragment frompNK2859, which carries the same kan gene. The results(data not shown) provided an unambiguous zbc location foreach insertion (Fig. 3b). We then used three-factor crosses tomap the rus locus relative to purE, adk, and the two zbcinsertions. The data (Table 2) gave the order adk-purE-zbc-2253-rus-zbc-2252, which places rus at about 12.5 minbetween coordinates 580 and 590 (Fig. 3b).

Suppression in rec' sbc' and recBC sbc strains. Althoughrus- was isolated in a recBC sbcBC background, it sup-presses muv mutations just as well in rec+ sbc+, recBC sbcA,and recBC sbcBC strains (Fig. 2b and data not shown). Italso suppresses ruvA60 in strain W3110 (data not shown),from which we conclude that suppression is not specific tothe AB1157 genetic background. We used a mvA60 recBC'sbc' strain to isolate a number of new suppressors byselecting for resistance to mitomycin. The suppressor waslinked topurE in every case tested (data not shown).The ruvA60 allele severely reduces the number of recom-

binants recovered in Hfr crosses with recBC sbcBC strainsand also causes abortive transfer of F-primes (3, 18). ms-i

J. BACTERIOL.

on Septem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

INDIRECT SUPPRESSION OF ruv MUTATIONS 4329

proCa.

Enis ruvA60,'uvA.

(Us

adK purE OmPT leuJ11

I - jI A I10 j 12 14

Iseo 580i800b. 5e0 58 600l ~ ~~~ ~ ~~I I

I I Il III11111 _ ~~~~~~~~~~~~~~~~~~~~~~I

l A. l l l l_ Ib. recesbc+ Ik i

zbc-2253-:*n zbc 2252:i*an0 20 40 60 FIG. 3. Map showing location of rus on the genetic map (2) (a)

and the Kohara et al. physical map (16) (b) of the E. coli chromo-\ ~o O some. The shaded restriction fragments in panel b are those identi-

fied by Southern analysis to be affected by the zbc-2252 and zbc-2253insertions. BamHI digests were not informative since the enzyme

(US(1N.V cuts the kan insertion used at both ends (15).

0 20 40 60 0 20 40 60

UV dose - J/m2FIG. 2. Suppression of UV sensitivity in ruv mutants. The

strains used are identified by genotype in each panel: (a) JC7623,RG108, N2238; (b) TNM759, N2057; (c) TNM733, TNM1124,GS1481; (d) TNM1115, N3684. The survival of the ms-i strainTNM733 (c) is almost identical to that of the rus' control, AB1157(data not shown).

eliminates these defects (Table 3a). It also eliminates theslight deficiency in recombination caused by ruvA60 in recksbc+ and recBC sbcA strains (Table 3b and data not shown).The rus mutation alone has no obvious effect on recombina-tion. Mutations in recA and recB reduce recombination byfactors of approximately 5,000-fold and 500-fold, respec-tively, in both rus ruv and nws ruv+ strains (data not shown),from which we deduce that recombination in a rus strainproceeds via a RecA- and RecBCD-dependent mechanism as

it does in the wild type. Mutations in recD, recF, rec, recN,recO, or recQ had no effect on recombination beyond thatseen in a rus' strain (data not shown).

Suppression of ruvB requires inactivation of ruvA. TheruvA200 allele, a nitrosoguanidine-induced point mutation(35), and ruvA59, another TnlO insertion (41), are sup-pressed just as well as ruvA60 (data not shown). However,ruvB mutations are suppressed rather poorly. We examinedseveral different alleles and obtained almost identical results(the data were similar to those obtained with ruvC strains[Table 3 and Fig. 2c]). The ability to fully suppress mutationsin ruvA but not those in ruvB was unexpected given that thegene products act together to provide a branch migrationactivity. Since the polar TnlO insertions are fully sup-pressed, RuvA may be limiting suppression in ruvB mutantsby preventing an alternative activity like RecG from gainingaccess to recombination intermediates.To test this possibility directly, we used a simple band

shift assay to examine binding of RecG to a syntheticX-junction in the presence of RuvA. We found that RuvAprevents binding even when RecG is in molar excess overRuvA (Fig. 4). In other experiments, we increased theRecG/RuvA ratio to 10:1. Again, the only complex detectedwas that formed by RuvA (data not shown). We alsoexamined dissociation of the junction. With 100 nM RecG inthe reaction mixture, inhibition was observed with as little as3.9 nM RuvA (Fig. 5, lane c). Dissociation was eliminatedwhen RuvA was increased to 15 nM (lane e). We concludethat binding of RuvA prevents RecG from gaining access to

TABLE 2. Mapping of rus by P1 transductional crosses

P1 donor Recipient No.a Marker segregation (% of total)"

N3603 (zbc-2252::kan) TNM716 (purE rus ruvA63) 160 42% purE rus', 52.4% pur+ rus, 5%purE nus, 0.6% pur+ rusN3605 (zbc-2253::kan) TNM716 (purE rus ruvA63) 148 11.5% purE nrs+, 47.3% pur+ rus+, 6.7%purE rus, 34.5% pur+ rusN3608 (ius-I purE zbc-2252) W3110 (wild type) 522 44% pur+ rus-i, 21%pur' n&s', 35% purE ms-i, 0.02%purE rus'N3603 (zbc-2252) TNM1031 (adk purE) 298 5%pur+ adk+, 50.6%pur+ adk, 1% purE adk+, 43.4%purE adk

a Selection was for Kmr transductants in each case.b Segregation at adk was monitored by growth at 42'C, segregation atpurE was monitored by the requirement for adenine, and segregation at rus was monitored

by sensitivity to mitomycin with W3110 as recipient and to both mitomycin and UV light with TNM716.

.1

.01

.001

.0001

0)a

._oI)c0

.1

.01

.001

.0001

VOL. 175, 1993

on Septem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

4330 MANDAL ET AL.

TABLE 3. Effect of rus-1 on DNA transfer and recombination in crosses with Hfr and F' donors

Relative yield of transconjugantsa

Recipient genotype Strain no. Viability x KL548 x GY2200" x KL227(F' Pro') A (TL+) (Pro')

rec+ ruv+ (control) AB1157 1.0 = 1.8 x 10' 1.0 = 2.6 x 107 1.0 = 1.4 x 107 1.0 = 1.7 x 107 1.0 = 1.1 X 107

(a) recBC sbcBCrus TNM734 0.32 1.12 1.70 0.40 1.41ruvA60 N2238 0.051 0.00048 0.75 0.0021 0.0065rus ruvA60 TNM844 0.24 1.1 1.77 0.41 1.09ruvC53 TNM839 0.012 0.00001 0.70 0.00025 0.00011rus ruvC53 TNM838 0.28 0.001 1.65 0.1 0.032ruvA60 AruvC64 TNM1120 0.188 0.00009 0.86 0.00086 0.0024rus ruvA60 AruvC64 TNM1206 0.37 0.64 2.4 0.35 0.84

(b) recBC+ sbcBC+rus TNM733 1.13 1.46 1.41 1.18 1.17ruvA60 N2057 0.48 0.88 1.07 0.24 0.44rusruvA60 TNM759 1.07 1.16 1.16 0.98 0.82recG258 N2731 0.69 1.16 1.16 0.31 0.39rus recG258 TNM883 0.52 1.10 0.86 0.10 0.16ruvA60 recG258 TNM896 0.10 0.20 0.83 0.00093 0.002rus ruvA60 recG258 TNM897 0.35 0.59 1.0 0.035 0.039AruvC64 GS1481 0.73 0.48 0.60 0.18 0.16rus AruvC64 TNM1124 1.34 0.98 0.89 0.64 0.75ruvA60 AruvC64 N3684 0.65 0.49 0.22 0.21 0.49rus ruvA60 AruvC64 TNM1115 0.78 0.76 1.09 0.80 0.86ruvA60 recG162 AM549 0.47 0.30 0.073 0.00073 0.0067rus AruvAC65 recGJ62 TNM1228 0.54 0.70 0.66 0.040 0.041AruvAC65 recG263 TNM1219 0.24 0.21 0.88 0.0013 0.002rus AruvAC65 recG263 TNM1215 0.5 0.52 1.18 0.042 0.038

a Under this spanner, the values in the first two columns provide different measures ofDNA transfer, while values in the second two columns provide measuresof recombination. Mating was for 30 (KL548), 40 (Hfr KL227), or 60 (Hfr GY2200) min. The transconjugant class selected is shown in parentheses. The absolutevalues shown for strain AB1157 are per milliliter of mating mixture. The data for the test strains are means of two to four experiments.

b A, plaques from zygotic induction of the Hfr X prophage; TL+, Thr+ Leu+.C The absolute value shown for strain AB1157 is per milliliter of recipient culture.

the junction. A similar situation could arise in vivo, espe-cially after SOS induction, and may explain why a multicopyruvA' plasmid makes ruv+ cells extremely sensitive to UVlight (36).

Suppression requires RecG. We considered the possibilitythat rus- might suppress ruvA(B) mutations by increasingexpression of RecG. To test this possibility, we introducedrecG' into a ruvA60 strain on high-copy-number plasmids(pBL125 and pBL135). There was no increase in resistanceto UV light with pBL125. We observed some increase withpBL135, where recG is expressed from the vector lacpromoter. However, the effect was very slight comparedwith that of ms-i and was the same with and without IPTG(isopropyl-0-D-thiogalactopyranoside) induction (data notshown). We also examined transcription from DNA frag-ments thought to contain the promoters that drive expres-sion of recG. Two large sections of the DNA upstream of therecG reading frame were cloned in front of the promoterlesslacZYA genes in pRS415. The fragment cloned in pTM101contains the putative P2 promoter; that in pTM102 containsboth P2 and P3 promoters (14, 24). These constructs wereintroduced into mus-1 and rus', A(pro-lac) strains, andsynthesis of P-galactosidase was measured relative to con-trols carrying pRS415. The data in Table 4 show that theregions cloned contain strong promoters. pTM102 producesslightly more 13-galactosidase than pTM101, which suggeststhat the putative P3 promoter in pTM102 is functional, assuggested previously (24). However, we observed no signif-

icant differences between the two strains with either of theplasmids made.These observations indicate that suppression is unlikely to

be due to increased synthesis of RecG. However, RecG isrequired for suppression. When we introduced a recG mu-tation into a rus ruvA60 strain, recombination was reducedsome 30-fold and sensitivity to UV light was increased to alevel typical of a ruv mutant (Table 3b and Fig. 6a). Weexamined a number of different recG alleles, including apoint mutation (recG162), an insertion (recG258), and adeletion (recG263). All three had the same effect. The loss ofrecG activity has a much greater effect than is seen in rus'rnv+ strains. However, the recG mus ruv strains are substan-tially more proficient in recombination and more resistant toUV light than the corresponding us' strains (Table 3b, Fig.6a, and data not shown) (17). These observations suggestthat mus-i activates some gene product to act with RecG inthe suppression of ruv mutations. If suppression was duesimply to increased expression of recG or to some activationof RecG protein, then mutation of recG would be expectedto have the same effect in both muv and mus nuv strains.

Suppression of ruvC. A possible clue as to the mode ofaction of ms-i came from studies on the suppression ofmuvC. We observed that ruvC mutations were suppressed tothe same intermediate extent as muvB mutations. Very sim-ilar results were obtained with muvC point mutations anddeletions (Fig. 2c, Table 3, and data not shown). To deter-mine whether this incomplete suppression of muvC was due

J. BACTERIOL.

on Septem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

INDIRECT SUPPRESSION OF ruv MUTATIONS

[RecG] nM 0 0 136 120v 25 50 100 0 12 12 12 12 12 12 12(5.15 gMjunction)[RuvA] nM 0 50 50_50 50 50 5050 _0_ 0 _15 31 62 125250500

a bocd ef gh i j k I

RuvA complex

RecG complex

junction

m n o p

FIG. 4. Effect of RuvA on binding of RecG to a synthetic X-junction. RecG at increasing concentrations and RuvA at a fixed concentration(lanes b to h) or vice versa (lanes j to p) were mixed on ice with 0.4 A.M 32P-5'-end-labelled X-junction DNA as described in Materials andMethods. Binding was initiated by the addition ofDNA to the reaction mixture. Lanes a and i are controls containing neither RuvA nor RecG.Complexes were separated on a low-ionic-strength gel at room temperature, and labelled DNA was detected by autoradiography.

to the negative effect of RuvA, as it was with ruvB, weconstructed a strain deficient for both ruvA and ruvC (seeMaterials and Methods) and introduced rus-i. The resultingstrain proved resistant to UV light and proficient in recom-bination (Fig. 2d and Table 3). Clearly, rus-i is able to fullysuppress iuvC provided that RuvA is inactive. Again, sup-pression depends on recG to the same extent it does in a -usruvA60 strain (Fig. 6b, Table 3, and data not shown). It isalso worth noting that the iuvA ruvC double mutants have

100 nM RecG

0 0 3.9 7.8 15 31 62 125 250 500 1000 nM RuvA

junction

,*06yl="

oligo 1+4

oligo 1+2

a b c d e f g h i j k

FIG. 5. Inhibitory effect of RuvA on dissociation of a syntheticX-junction by RecG. Reaction mixtures containing 0.15 ,uM 5'-32P-labelled junction DNA and RecG and RuvA at the indicated con-centrations were incubated for 30 min at 37°C. Reactions were

stopped and deproteinized as described in Materials and Methods,and the products were analyzed on a 10% polyacrylamide gel.

the same phenotype as the single mutants, no more and noless.The ability to suppress a deficiency in ruvC in a way that

depends on recG is highly significant. There is no evidencethat RecG has a nuclease activity that can cleave Hollidayjunctions. We assume, therefore, that suppression dependsnot only on RecG but also on the activation perhaps of analternative resolvase to replace RuvC. This is consistentwith the finding that rus-i makes a ruvC recG strain moreresistant to UV light (Fig. 6b).

DISCUSSION

We have shown that ruv mutations can be suppressedindirectly by an additional mutation at a new locus calledrus. With ruvA strains, the defects in both recombinationand DNA repair are suppressed completely, but with ruvBand ruvC strains, suppression is incomplete unless ruvA isalso eliminated. We found that RuvA prevents RecG fromgaining access to Holliday junctions in vitro and suspecttherefore that it can also interfere with the processing ofthese structures in vivo when either RuvB or RuvC isabsent. This conclusion is consistent with the finding thatsuppression is largely eliminated by mutation of recG.

TABLE 4. Synthesis of 3-galactosidase in rus' and ius-i strainscarrying fusions of the recG promoter region to lacZ

1-Galactosidase units/A650 UStrain Genotype pRS415 pTM101 pTM102

(vector) (P2) (P2 + P3)

TNM1107 rus+ 4 1,956 2,423TNM1106 -us-l 5 2,009 2,766

A.& -. iA AA AWL-. Iw

4331VOL. 175, 1993

on Septem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

4332 MANDAL ET AL.

1 .ocbo usdruvAC65 with other proteins (5, 6). The idea is particularly appealing.O7 a-.NruvA60 us since it explains why rus mutations are able to suppress both

ruvAB and ruvC mutations. It also accounts neatly for theIrecG325B>°z fact that combining recG and ruv mutations leads to in-

recG263 creases in sensitivity to UV light and decreases in theItAz efficiency of recombination that are essentially identical in

the case of all three of the ruv genes (17, 25). Accordingly,the synergism observed in ruv recG strains would be due tothe combined effects of eliminating both branch migration

ruvA60rus rusAUvAC65 and resolution of Holliday junctions.1re58 recG263 In studies to be reported elsewhere, we have shown thatthe effect of a rus mutation can be mimicked by a multicopyplasmid harboring the rus+ region of the chromosome. Wesuspect, therefore, that the rus mutation leads to the in-

ruvA60 recG258 AruvAC65re"Mcreased expression of an activity that helps to process

ruvA6O rec0256 AruvAC65recG263 Holliday junction intermediates in association with RecG.>1a. , . , . . b. , . , . . Whether this activity catalyzes symmetrical cleavage of0 20 40 60 0 20 40 60 junctions like RuvC remains to be determined. We cannot

UVdoseJ/m2 rule out the possibility that the rus mutation provides aUV dose

of

2 radically different recombination mechanism. To account for6. Effect of recG on suppression of ruv mutations. The the phenotype of ruv mutants, we assume that the rusused are identified by genotype in each panel: (a) N2731, activity is normally expressed sufficiently to promote ex-6, TNM897; (b) TNM1072, TNM12O8, TNM1215, TNM1219. atvt snral xrse ufcetyt rmt x

-a for us-1 ruvA60 from Fig. 2b are included for comparison changes in genetic crosses but cannot cope with the in-1a (dashed line), creased demands of repair. This hypothesis has a number of

clear predictions which we are currently testing. A molecu-lar analysis of the rus region is also under way and shouldreveal exactly how the suppression of ruv is achieved.

requirement for RecG fits with the ability of this A further corollary to this work is that the UV sensitivityto catalyze branch migration of Holliday junctions in of a ruv mutant provides a measure of the part played by

25, 51). Presumably, RecG replaces RuvAB. How- Holliday junction resolution in the repair of damaged DNA.his immediately raises the question of why ruvA and It follows that the increase in sensitivity seen when recG isrutants are normally very sensitive to UV light. If also inactivated provides a measure of the part played bycan replace RuvAB, they should be resistant. The branch migration. From the data presented here and inocus is rather poorly expressed, and it may be that earlier studies, it would appear that branch migration ands insufficient RecG in a ruse strain to compensate for resolution are both very important. They support the modelthe SOS-inducible RuvAB proteins (24). If true, then presented in Fig. 1. We propose that after DNA replicationutation might achieve its effect by increasing the level has left a gap in the newly synthesized DNA opposite theG. There are several reasons why this explanation is lesion in the template strand, RecA protein binds to the gapoly incorrect. First, we found that multicopy recG' and initiates strand exchange as suggested initially by Ruppds fail to produce a substantial increase in the survival et al. (33) and West et al. (50). This leads to a Holliday-irradiated ruvAB mutants. Second, us-i does not junction that can be branch migrated past the UV lesion byto increase the activity of recG promoters. Third, RuvAB or RecG, allowing Uvr(A)BC excinuclease to re-

is very much more active than RuvAB in terms of its move the dimer. After filling in of the excision tract, repairto dissociate synthetic junctions (25, 26). The branch can now be completed in one of two ways. Either RuvCion activity of RecG is likely to compare very favor- resolves the Holliday junction by cleavage, or, alternatively,herefore, with that of RuvAB, even in SOS-induced RuvAB or RecG catalyzes branch migration in the reverseells. direction to return the exchanged strands to their original-cG or RuvAB can independently satisfy most of the partners. We propose that resolution and reverse branchieed for a branch migration activity, it follows that migration are independent routes to repair. This followscn of ruvA and ruvB must have another effect on from the observation that eliminating both activities byto account for the sensitivity to UV light. nrv single combining ruv and recG mutations has a synergistic effect onts have essentially identical phenotypes regardless of sensitivity to UV light. A rus ruv recG strain would perhapsof the three genes is affected, and as we have remain able to resolve Holliday junctions made by RecA, butstrated there is no further decrease in recombination branch migration would be restricted to that catalyzed byir when mutations eliminating both RuvAB and RuvC RecA itself.

are combined. The identical phenotype and lack of additivitymay reflect the fact that RuvAB and RuvC catalyze consec-utive steps in the same repair process. However, given theoverlap between RecG and RuvAB, another possibility isthat ruvA, ruvB, and ruvC mutants are all deficient inresolution of Holliday junctions, with RecG providing all ormost of the activity needed to promote branch migration.This situation would arise if RuvC were able to cleavejunctions in vivo only in the presence of RuvAB. Theobservation that RuvC functions extremely inefficiently invitro is probably indicative of the fact that it normally works

ACKNOWLEDGMENTSWe thank Matthew Whitby and Lizanne Ryder for their support

and encouragement and Carol Buckman, Lynda Harris, and LisaCorbett for excellent technical assistance. Strain RG108 was iso-lated by Richard Griffiths during the course of an undergraduateproject in the laboratory.

This work was supported by grants to R.G.L. from the Scienceand Engineering Research Council, the Medical Research Council,and the Wellcome Trust. T.N.M. was supported by an SERCresearch studentship, and G.J.S. was supported by a Royal SocietyUniversity Research Fellowship.

CD.c

C>of

COc

L. .000

.0000

.0000,

FIG.strainsTNM89The datin panel

Theproteinvitro ('ever, truvB nRecGrecG kthere i,loss ofa rus mof Reckprobabplasmi(of UVappearRecG iabilitymigratiably, tlrus+ cc

If Recell's rmutati(repairmutantwhichdemon;or repa

J. BACTERIOL.

on Septem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

INDIRECT SUPPRESSION OF ruv MUTATIONS 4333

REFERENCES1. Bachmann, B. J. 1987. Derivations and genotypes of some

mutant derivatives of Escherichia coli K-12, p. 1190-1219. InF. C. Neidhardt, J. L. Ingraham, K. B. Low, B. Magasanik, M.Schaechter, and H. E. Umbarger (ed.), Escherichia coli andSalmonella typhimurinum: cellular and molecular biology. Amer-ican Society for Microbiology, Washington, D.C.

2. Bachmann, B. J. 1990. Linkage map of Escherichia coli K-12,edition 8. Microbiol. Rev. 54:130-197.

3. Benson, F., S. Colier, and R. G. Lloyd. 1991. Evidence ofabortive recombination in ruv mutants of Escherichia coli K-12.Mol. Gen. Genet. 225:266-272.

4. Benson, F. E., G. T. Illing, G. J. Sharples, and R. G. Lloyd.1988. Nucleotide sequencing of the ruv region of Escherichiacoli K-12 reveals a LexA regulated operon encoding two genes.Nucleic Acids Res. 16:1541-1549.

5. Connolly, B., C. Parsons, F. E. Benson, H. J. Dunderdale, G. J.Sharples, R. G. Lloyd, and S. C. West. 1991. Resolution ofHolliday junctions in vitro requires Escherichia coli ruvC geneproduct. Proc. Natl. Acad. Sci. USA 88:6063-067.

6. Dunderdale, H. J., F. E. Benson, C. A. Parsons, G. J. Sharples,R. G. Lloyd, and S. C. West. 1991. Formation and resolution ofrecombination intermediates by E. coli RecA and RuvC pro-teins. Nature (London) 354:506-510.

7. Echols, H., and M. F. Goodman. 1991. Fidelity mechanisms inDNA replication. Annu. Rev. Biochem. 60:477-511.

8. Holliday, R. 1964. A mechanism for gene conversion in fungi.Genet. Res. Camb. 5:282-304.

9. Howard-Flanders, P., L. Theriot, and J. B. Stedeford. 1969.Some properties of excision-defective recombination-deficientmutants of Escherichia coli K-12. J. Bacteriol. 97:1134-1141.

10. Howard-Flanders, P., S. C. West, and A. J. Stasiak. 1984. Roleof RecA spiral filaments in genetic recombination. Nature(London) 309:215-220.

11. Iwasaki, H., T. Shiba, K. Makino, A. Nakata, and H. Shinagawa.1989. Overproduction, purification, and ATPase activity of theEscherichia coli RuvB protein involved in DNA repair. J.Bacteriol. 171:5276-5280.

12. Iwasaki, H., M. Takahagi, A. Nakata, and H. Shinagawa. 1992.Escherichia coli RuvA and RuvB proteins specifically interactwith Holliday junctions and promote branch migration. GenesDev. 6:2214-2220.

13. Iwasaki, H., M. Takahagi, T. Shiba, A. Nakata, and H. Shina-gawa. 1991. E. coli RuvC protein is an endonuclease thatresolves the Holliday structure, an intermediate of homologousrecombination. EMBO J. 10:4381-4389.

14. Kalman, M., H. Murphy, and M. Cashel. 1992. The nucleotidesequence of recG, the distal spo operon gene in Escherichia coliK-12. Gene 110:95-99.

15. Kleckner, N., J. Bender, and S. Gottesman. 1991. Uses oftransposons with emphasis on TnlO. Methods Enzymol. 204:139-180.

16. Kohara, Y., K. Akiyama, and K. Isono. 1987. The physical mapof the whole E. coli chromosome: application of a new strategyfor rapid analysis and sorting of a large genomic library. Cell50:495-508.

17. Lloyd, R. G. 1991. Conjugational recombination in resolvase-deficient ruvC mutants of Escherchia coli K-12 depends onrecG. J. Bacteriol. 173:5414-5418.

18. Lloyd, R. G., F. E. Benson, and C. E. Shurvinton. 1984. Effectof ruv mutations on recombination and DNA repair in Esche-richia coli K12. Mol. Gen. Genet. 194:303-309.

19. Lloyd, R. G., and C. Bucknan. 1985. Identification and geneticanalysis of sbcC mutations in commonly used recBC sbcBstrains of Eschenichia coli K-12. J. Bacteriol. 164:836-844.

20. Lloyd, R. G., and C. Buckman. 1991. Genetic analysis of therecG locus of Escherichia coli K-12 and of its role in recombi-nation and DNA repair. J. Bacteriol. 173:1004-1011.

21. Lloyd, R. G., N. P. Evans, and C. Buckman. 1987. Formation ofrecombinant lacZ+DNA in conjugational crosses with a recBmutant of Escherichia coli K12 depends on recF, recJ, andrecO. Mol. Gen. Genet. 209:135-141.

22. Lloyd, R G., B. Low, G. N. Godson, and E. A. Birge. 1974.

Isolation and characterization of an Escherichia coli K-12mutant with a temperature-sensitive RecA- phenotype. J. Bac-teriol. 120:407-415.

23. Lloyd, R. G., M. C. Porton, and C. Buckman. 1988. Effect ofrecF, recJ, recN, recO, and ruv mutations on ultraviolet sur-vival and genetic recombination in a recD strain of Escherichiacoli K-12. Mol. Gen. Genet. 212:317-324.

24. Lloyd, R. G., and G. J. Sharples. 1991. Molecular organizationand nucleotide sequence of the recG locus of Escherichia coliK-12. J. Bacteriol. 173:6837-6843.

25. Lloyd, R. G., and G. J. Sharples. 1993. Dissociation of syntheticHolliday junctions by E. coli RecG protein. EMBO J. 12:17-22.

26. Lloyd, R. G., and G. J. Sharples. 1993. Processing of syntheticHolliday junctions by RuvAB and RecG proteins of Escherzchiacoli K-12. Nucleic Acids Res. 21:1719-1725.

27. Miller, J. H. 1972. Experiments in molecular genetics. ColdSpring Harbor Laboratory, Cold Spring Harbor, N.Y.

28. Oden, K. L., L. C. De Veaux, C. R. T. Vibat, J. E. Cronan, Jr.,and R. B. Gennis. 1990. Genomic replacement in Eschenchiacoli K-12 using covalently closed circular plasmid DNA. Gene96:29-36.

29. Otsuji, N., H. Iyehara, and Y. Hideshima. 1974. Isolation andcharacterization of an Escherichia coli ruv mutant which formsnonseptate filaments after low doses of ultraviolet light irradia-tion. J. Bacteriol. 117:337-344.

30. Parsons, C. A., B. Kemper, and S. C. West. 1990. Interaction ofa four-way junction in DNA with T4 endonuclease VII. J. Biol.Chem. 265:9285-9289.

31. Parsons, C. A., I. Tsaneva, R. G. Lloyd, and S. C. West. 1992.Interaction of E. coli RuvA and RuvB proteins with syntheticHolliday junctions. Proc. Natl. Acad. Sci. USA 89:5452-5456.

32. Rupp, W. D., and P. Howard-Flanders. 1968. Discontinuities inthe DNA synthesized in an excision-defective strain of Esche-nichia coli following ultraviolet irradiation. J. Mol. Biol. 31:291-304.

33. Rupp, W. D., C. E. Wilde, D. L. Reno, and P. Howard-Flanders.1971. Exchanges between DNA strands in ultraviolet-irradiatedEscherichia coli. J. Mol. Biol. 61:25-44.

34. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecularcloning: a laboratory manual. Cold Spring Harbor Laboratory,Cold Spring Harbor, N.Y.

35. Sargentini, N. J., and K. C. Smith. 1989. Role of ruvAB genes inUV- and y-radiation and chemical mutagenesis in Escherichiacoli. Mutat. Res. 215:115-129.

36. Sharples, G. J., F. E. Benson, G. T. Illing, and R. G. Lloyd.1990. Molecular and functional analysis of the ruv region ofEscherichia coli K-12 reveals three genes involved in DNArepair and recombination. Mol. Gen. Genet. 221:219-226.

37. Sharples, G. J., and R. G. Lloyd. 1991. Resolution of Hollidayjunctions in Escherichia coli: identification of the ruvC geneproduct as a 19-kilodalton protein. J. Bacteriol. 173:7711-7715.

38. Shiba, T., H. Iwasaki, A. Nakata, and H. Shinagawa. 1991.SOS-inducible DNA repair proteins, RuvA and RuvB, of Esch-enchia coli: functional interactions between RuvA and RuvBfor ATP hydrolysis and renaturation of the cruciform structurein supercoiled DNA. Proc. Natl. Acad. Sci. USA 88:8445-8449.

39. Shinagawa, H., K. Makino, M. Amemura, S. Kimura, H.Iwasaki, and A. Nakata. 1988. Structure and regulation of theEscherichia coli ruv operon involved in DNA repair and recom-bination. J. Bacteriol. 170:4322-4329.

40. Shurvinton, C. E., and R. G. Lloyd. 1982. Damage to DNAinduces expression of the ruv gene of Eschenchia coli. Mol.Gen. Genet. 185:352-355.

41. Shurvinton, C. E., R. G. Lloyd, F. E. Benson, and P. V. Attfield.1984. Genetic analysis and molecular cloning of the Escherichiacoli ruv gene. Mol. Gen. Genet. 194:322-329.

42. Silhavy, T. J., M. L. Berman, and L. W. Enquist. 1984.Experiments with gene fusions. Cold Spring Harbor LaboratoryPress, Cold Spring Harbor, N.Y.

43. Simons, R. W., F. Houman, and N. Kleckner. 1987. Improvedsingle and multicopy lac-based cloning vectors for protein andoperon fusions. Gene 53:85-96.

44. Tsaneva, I., G. Illing, R. G. Lloyd, and S. C. West. 1992.

VOL. 175, 1993

on Septem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

4334 MANDAL ET AL.

Purification and physical properties of the RuvA and RuvBproteins of Escherichia coli. Mol. Gen. Genet. 235:1-10.

45. Tsaneva, I. R., B. Muller, and S. C. West. 1992. ATP-dependentbranch migration of Holliday junctions promoted by the RuvAand RuvB proteins of E. coli. Cell 69:1171-1180.

46. Van Houten, B. 1990. Nucleotide excision repair. Microbiol.Rev. 54:18-51.

47. Way, J. C., M. A. Davis, D. Morisato, D. E. Roberts, and N.Kleckner. 1984. New TnlO derivatives for construction of lacZoperon fusions by transposition. Gene 32:369-379.

48. West, S. C. 1992. Enzymes and molecular mechanisms ofgenetic recombination. Annu. Rev. Biochem. 61:603-640.

49. West, S. C., E. Cassuto, and P. Howard-Flanders. 1981. Homol-ogous pairing can occur before DNA strand separation ingeneral genetic recombination. Nature (London) 290:29-33.

50. West, S. C., E. Cassuto, and P. Howard-Flanders. 1981. Mech-anism of E. coli RecA protein directed strand exchanges inpost-replication repair of DNA. Nature (London) 294:659-662.

51. Whitby, M. C., L. Ryder, and R. G. Lloyd. Unpublished results.

J. BACrERIOL.

on Septem

ber 14, 2020 by guesthttp://jb.asm

.org/D

ownloaded from